Research map / Label-efficient and robust learning / Recognition and long-tail
NeuroNCAP: Photorealistic Closed-loop Safety Testing for Autonomous Driving
William Ljungbergh, Adam Tonderski, Joakim Johnander, Holger Caesar, Kalle Åström, Michael Felsberg, Christoffer Petersson
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| apply UniAD's collision-avoidance trajectory post-processing in closed loop (ablation) | UniAD · NeuroNCAP stationary | collision rate (%): 87.80 -> 34.80 | Improved |
| add the same post-processing to VAD (ablation) | VAD · NeuroNCAP (average) | NeuroNCAP score: 0.66 -> 2.75 | Improved |
| add post-processing on frontal scenarios for UniAD (ablation) | UniAD · NeuroNCAP frontal | collision rate (%): 98.40 -> 92.40 | Improved |
| replace end-to-end planning with naive braking rule on perception outputs | UniAD/VAD perception · NeuroNCAP stationary | collision rate (%): 9.60 and 6.00 vs 87.80 and 96.20 | Improved |
| naive braking rule in frontal and side scenarios | naive baseline · NeuroNCAP frontal and side | collision rate (%): 100.00 collision rate | No clear effect |
| render images with NeRF instead of real images (open-loop) (ablation) | UniAD · nuScenes 14 sequences | NDS: 0.490 -> 0.489 | No clear effect |
| render images with NeRF instead of real images (open-loop) (ablation) | VAD · nuScenes 14 sequences | NDS: 0.449 -> 0.413 | Hurt |
| render images with NeRF instead of real images (open-loop) (ablation) | UniAD and VAD · nuScenes 14 sequences | ADE at 3.0s: 1.16 -> 1.24 (UniAD); 1.01 -> 1.16 (VAD) | Mixed |
Closest papers by meaning
- NAVSIM: Data-Driven Non-Reactive Autonomous Vehicle Simulation and Benchmarking (NeurIPS 2024 · similarity 0.85)
- NeuRAD: Neural Rendering for Autonomous Driving (CVPR 2024 · similarity 0.85)
- Bench2Drive: Towards Multi-Ability Benchmarking of Closed-Loop End-To-End Autonomous Driving (NeurIPS 2024 · similarity 0.85)
- DriveHierarchy: A Benchmark for Diagnosing VLM Driving Capabilities from Open-Loop Understanding to Closed-Loop Execution (NeurIPS 2026 · similarity 0.85)
- DrivingSphere: Building a High-fidelity 4D World for Closed-loop Simulation (CVPR 2025 · similarity 0.84)
- DiffScene: Diffusion-Based Safety-Critical Scenario Generation for Autonomous Vehicles (AAAI 2025 · similarity 0.84)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
